BACKGROUND AND OBJECTIVE:The automated recognition of surgical phases in intraoperative videos represents a critical milestone in the digital transformation of surgery. It forms the foundation for surgical assistance systems and enhanced decision support, contributing to increased safety, efficiency, and precision in surgical procedures. Traditional deep learning methods often fall short in this domain due to their dependency on extensive annotated datasets, which are challenging to obtain in medical contexts due to privacy concerns and data scarcity. This study explores the potential of few-shot learning as a paradigm for overcoming data limitations in surgical phase recognition. METHODS:By leveraging the ability to generalize from minimal examples, a transformer-based few-shot learning (FSL) model for action recognition was adapted to the recognition of surgical phases using the Cholec80 dataset, which consists of videos from cholecystectomy surgeries. The model's performance was evaluated across three experimental splits to assess domain-specific and cross-domain performance: Split 1, where the model was trained on a surgical dataset and tested on Cholec80; Split 2, which introduced variations in surgical environments; and Split 3, where the model was trained on action recognition data and tested on surgical data. RESULTS:The model achieved test accuracies of 89.0%, 75.4%, and 49.1% in these splits, respectively. While FSL demonstrates strong applicability to surgical data, domain-specific training remains crucial for optimal performance. Notably, these results were obtained using only a few labeled support examples per phase, illustrating the data efficiency of the approach. CONCLUSION:This study provides an initial foundation for applying few-shot learning to surgical phase recognition and demonstrates its feasibility under low-label and transfer settings. While domain-specific training remains important, the results indicate that FSL is a promising direction for surgical workflow analysis when annotation resources are limited.

Soft Everting Robots (SER) are a subclass of soft robotic systems that move by body eversion, enabling highly compliant and adaptive locomotion. These properties make them attractive for medical use, particularly in endoluminal procedures such as colonoscopy or vascular navigation. A structured literature review was performed following the PRISMA methodology. In total, 50 publications were identified that explicitly investigated SER in medical contexts. The publications were categorized by application area, technical design aspects, and reported challenges. Recurring issues include safe interaction with delicate tissue, prevention of leakage, miniaturization to anatomical constraints, sterility and reusability concepts, and reliable navigation in tortuous pathways. SER are additionally compared against related technologies - as these often surface in SER-related searches and can be confused with SER approaches - and the commercialization landscape is briefly outlined. By consolidating these findings, the review provides a structured overview of the state of the art and outlines guidelines for design, control, and the potential future clinical implementation of SER.

Evaluation metrics are essential for assessing the performance of AI models and enabling comparability across different approaches. However, in surgical phase recognition, their application remains inconsistent. This study provides a systematic analysis of evaluation metrics used in this field, aiming to support standardization and encourage more explicit explanation of metric selection. A systematic literature review was conducted following the PRISMA framework, covering publications from 2016 to 2025. In total, 50 studies were identified and analyzed with respect to the evaluation metrics applied, with particular focus on relaxed boundaries and the distinction between online and offline models. The results show that accuracy, precision, and recall are the most frequently used metrics, followed by the jaccard index. Since around 2023, an increasing use of segment-based metrics can be observed, reflecting a growing emphasis on temporal dynamics. No significant differences in metric selection between online and offline approaches were identified. Furthermore, many studies do not provide explicit explanation for their choice of evaluation metrics. Instead, they often rely on commonly used metrics or practices adopted from previous work. Additionally, results obtained with relaxed boundaries tend to exhibit higher standard deviations compared to those without relaxed boundaries. Based on these findings, it is recommended to more explicitly align evaluation metrics with model objectives and to further investigate the distinction between online and offline settings. Adaptations of the f1-score, such as the fβ-score, may provide a more flexible evaluation framework. Furthermore, the Matthews Correlation Coefficient represents a promising complementary metric, particularly for multi-class settings, if appropriately normalized. Future work should focus on clearly explaining metric selection and consistently reporting relaxed boundaries to improve transparency, comparability, and interpretability in surgical phase recognition.

A recent focus has been on developing wearable health solutions that allow users to seamlessly track their health metrics during their daily activities, providing convenient and continuous access to vital physiological data. This work investigates a heart rate (HR) monitoring system and compares the HR measurement from two potential sites for foot wearable technologies. The proposed system used a commercially available photoplethysmography sensor (PPG), microcontroller, Bluetooth module, and mobile phone application. HR measurements were obtained from two anatomical sites, i.e., the dorsalis pedis artery (DPA) and the posterior tibial artery (PTA), and compared to readings from the Apple Smartwatch during standing and walking tasks. The system was validated on twenty healthy volunteers, employing ANOVA and Bland-Altman analysis to assess the accuracy and consistency of the HR measurements. During the standing test, the Bland-Altman analysis showed a mean difference of 0.08 bpm for the DPA compared to a smaller mean difference of 0.069 bpm for the PTA. On the other hand, the walking test showed a mean difference of 0.255 bpm and -0.06 bpm for the DPA and PTA, respectively. These results showed a high level of agreement between the HR measurements collected at the foot with the smartwatch measurements, with superiority for the HR measurements collected at the PTA.

BackgroundThe ergonomics of flexible endoscopes require improvement as the current design carries a high risk of musculoskeletal injury for endoscopists. Robotic systems offer a solution by separating the endoscope from the control handle, allowing a focus on ergonomics and usability. Despite the increasing interest in this field, little attention has been paid towards developing ergonomic human input devices. This study addresses two key questions: How can handheld control devices for flexible robotic endoscopy be designed to prioritize ergonomics and usability? And, how effective are these new devices in a simulated clinical environment?MethodsAddressing this gap, the study proposes two handheld input device models for controlling a flexible endoscope in four degrees of freedom (DOFs) and an endoscopic instrument in three DOFs. A two-stage evaluation was conducted with six endoscopists evaluating the physical ergonomics and a final clinical user evaluation with seven endoscopists using a virtual colonoscopy simulator with proportional velocity and position mapping.Results and discussionBoth models demonstrated clinical suitability, with the first model scoring 4.8 and the second model scoring 5.2 out of 6 in the final evaluation. In sum, the study presents two designs of ergonomic control devices for robotic colonoscopy, which have the potential to reduce endoscopy-related injuries. Furthermore, the proposed colonoscopy simulator is useful to evaluate the benefits of different mapping modes. This could help to optimize the design and control mechanism of future control devices.

In the study of Activities of Daily Living (ADLs), limb coordination is essential. Specifically, upper extremity bimanual tasks significantly influence human capabilities. Conditions such as nerve injuries or strokes can result in unilateral paralysis. Rehabilitative assistive devices aim to mitigate these functional impairments. This paper demonstrates the importance of bimanual tasks and their significance in developing assistive devices. Six bimanual ADLs were defined, encompassing both standing and seated tasks, various categories of ADLs, and different workspace areas. While a laboratory setting inherently has limitations in replicating realistic scenarios, representative movement patterns of daily living were identified.

Objectives The objective of this study was to develop and characterize a novel low-cost, flexible sensor system for ground reaction force (GRF) measurements for biomedical applications. The system aims to provide GRF measurements across customizable areas up to 2 m 2 , suitable for integration into various medical and rehabilitation devices. Methods The sensor system was constructed using multiple discrete resistive sensor modules. Each module had a quadratic shape and an edge length of 7.5 cm. The system utilized ESD packing-foam as resistive sensing material and conductive textile as electrodes. Measurements were conducted using an Arduino Nano microcontroller, a Wheatstone bridge circuit and analogue multiplexers. A demonstrator, integrating the sensor modules in a sports mat was built to show the functionality. Results The proposed system was capable of measuring forces up to 330 N. The sensor modules have an exponential force-resistance characteristic curve and showed inter-module and inter-day variability in the range of commercially available sensor systems’ accuracy. The demonstrator enabled to visualize changes in weight distribution on its surface. Conclusions The developed sensor system offers a reliable, flexible, and low-cost solution for GRF analysis in biomedical applications, providing data e.g. for rehabilitation feedback.

Each year, thousands of individuals, particularly young adults, experience traumatic brachial plexus injuries (TBPIs), leading to significant limitations, permanent disabilities, reduced quality of life, and infrequent return to work. Current treatments and assistive devices have shown limited success, resulting in considerable social and economic challenges for patients. Given the devastating nature of this injury and the lack of literature on return to work rates among young adults, this study aims to determine the percentage of individuals reintegrating into work after a TBPI. Furthermore, it compares outcomes across different health care systems, including those in Germany, Serbia, and the United Kingdom. This dual approach has been selected to investigate the influence of various factors on the outcomes associated with returning to work after TBPI. Preliminary findings indicate that approximately 60% of patients with TBPI return to work, although most require a change in their occupational roles. Despite variations in health care systems and governmental support, the reintegration of patients with TBPI into work and society remains a critical and universal challenge. This comparative analysis highlights disparities in TBPI research and outcomes, providing valuable insights for future improvements in patient care and support mechanisms.

In order to improve minimally invasive surgical procedures, efforts are being made to miniaturise existing flexible, cableactuated instruments and entire endoscopes. Concentric Tube Robots (CTR) are considered a promising solution in this context, as they enable high precision with minimal construction volume. However, their practical use has so far been limited by considerable technical requirements – in particular with regard to complex path planning to avoid snapping effects and the need for robotic control due to non-independently controllable degrees of freedom. Against this background, the approach presented aims to make CTR more accessible through targeted technical modifications and to significantly reduce the infrastructure costs. This could not only facilitate integration into existing surgical systems, but also promote the clinical dissemination of this technology. This work investigates the torsional behaviour of CTR, in which magnetic guides were implemented along the tubes. The goal was to reduce torsion in the CTR prototypes and enable more stable discrete positions for the CTR tip by using magnetic guides on the inner and outer tubes. For validation, two prototypes with different magnetic configurations were developed: the 6x1 model and the 5x3 model. These setups were tested in comparison to models in which the outer tubes had no magnetic guides, so no magnetic interaction occurred between the inner and outer tubes. First, suitable polymeric magnetic materials were selected and integrated into polyamide tubes. The validation experiments involved a 720Grad rotation in one direction followed by another 720Grad in the opposite direction, during which the torque at the base of each inner tube and the relative twist angle between the bases and tips of each tube pair were measured. Firstly, it became clear that the torsion effects were influenced by the magnetic guides. Nevertheless, torsion still occurred in the magnetic models, indicating that the magnetic stabilization was insufficient to fully suppress the torsion. Furthermore, the results showed that the 6x1 model exhibited a slight reduction in twist angle and therefore in torsion compared to its control model, while the 5x3 model experienced greater twist. Additionally, the rotation recordings of the tips revealed that neither the 6x1 nor the 5x3 model was able to maintain discrete positions. Despite the presence of magnetic guides, the twist in both models remained continuous. Thus, the desired positions were not stabilized under all circumstances. Future development will investigate the use of segmented magnetic structures, miniaturisation of the mechanical setup, the inclusion of working channels, and the implementation of a model-based control algorithm.

Capsule endoscopy represents an appealing alternative to conventional endoscopic procedures that cause discomfort to the patient and are considered tedious to the operator due to difficulty of endoscope manipulation because of the limited degrees of freedom (DOF). Therefore, active capsule endoscopy has become an important research topic as it enables targeted intestinal examination by controlling the capsule’s movements inside the gastrointestinal (GI) tract. An overview of the current state of art of active capsule endoscope systems is provided in this paper, discussing their different locomotion methods that can be classified to internally actuated and externally actuated. In addition to providing an overview of capsule endoscope functions, since the requirements of capsule endoscopy actuation (passive or active locomotion) might derive from the purpose/ function of the capsule. Developed capsule endoscope systems are discussed and systems with similar locomotion methods are compared. Generalized control schemes are presented to demonstrate the general control strategies implemented in each group of active capsule endoscope systems.

In this paper, different welding techniques for manufacturing pneumatic artificial muscles (PAMs) for the use in soft everting robots for colonoscopy are presented. The focus is on developing reproducible prototyping methods for manufacturing robust and cost-efficient PAMs. Two welding techniques, including laser welding and contact welding, are analyzed for their effectiveness in creating durable and scalable PAMs. Preliminary results show that laser welding offers high precision but inconsistent airtightness, crucial for PAM functionality. In contrast, contact welding, though slower, provides more reliable airtight seals and better reproducibility. Optimized contact welding settings balance weld seam thickness and production time. Further refinements are needed to improve speed and seam quality, with potential for large-scale production. These findings offer valuable insights for scalable PAM manufacturing in soft robotics.

Cable-Driven Parallel Robots (CDPR) have gained increasing attention in human-machine-interaction, particularly in haptics and rehabilitation. Despite their different objectives, both fields share fundamental requirements such as safety, configurability and precise force control. This review provides an overview of existing CDPR implementations designed for human interaction, categorizing and analyzing systems based on their design principles and kinematic properties. Key motivations for using CDPR in these applications are discussed, along with current limitations and promising directions. The findings indicate that CDPR offer an underrated yet promising solution for human-machine-interaction.

The increasing use of single-use flexible endoscopes raises concerns about their environmental impact. This systematic review includes 18 studies comparing singleuse and reusable endoscopes based on quantitative analyses on their environmental impact. Types included are bronchoscopes, cystoscopes, duodenoscopes, gastroscopes, laryngoscopes, and ureteroscopes. The number of available studies differs by endoscope type, with cystoscopes and bronchoscopes being most frequently assessed (seven and four studies). Findings indicate that the environmental impact depends on the specific endoscope type, with single-use cystoscopes being favorable, while for duodenoscopes, data suggest the reusable option to be more sustainable. However, results are highly study-dependent, influenced by regional circumstances such as transport distances, waste management, and reprocessing assumptions. Further quantitative data from underrepresented regions could help close remaining knowledge gaps for gastroscopes and laryngoscopes, and innovative materialsaving technologies may help reduce the carbon footprint of disposable devices where indispensable.

BACKGROUND:The examination of lymphogenic metastasis is the basis for the treatment of tumors. It primarily targets the sentinel lymph nodes (SLN). Common imaging techniques to identify SLNs expose patients to high doses of radiation. Ultrasound imaging supported by radiation detectors is a more patient-friendly alternative that can be used intraoperatively. This work aims to develop a gamma source detector which can localize the source and can be combined with a commercial ultrasound system to a dual-modality system. METHODS:The problem is modeled, and a localization algorithm is developed that robustly determines position estimates from noisy sensor readings. An accuracy optimized arrangement of sensors is determined. The concept is verified by using a test setup based on infrared radiation. Furthermore first tests with gamma radiation are performed. RESULTS:An infrared radiation source with unknown radiant flux is localized with a spatial deviation of 3.64 ± 1.66 mm (2σ) on average within a wide measurement range of 40 × 20 × 60 mm3. A visualization setup to combine and display data from the developed sensor system and an ultrasound system is shown, that achieves a display rate of 3 frames/s. Tests with gamma radiation show that two gamma source positions can be distinguished. CONCLUSIONS:The system presented allows 3D localization of a gamma radiation source with reduced radiation exposure to patients. The measurements are combined registration-free with ultrasound imaging to a dual-modality system and provide an enriched database for SLN identification. To reduce measurement times, more efficient sensors must be used.

PurposeAutomatic recognition of surgical workflows is a complex yet essential task of context-aware systems in the operating room. However, achieving high accuracy in phase recognition remains a challenge due to the complexity of surgical procedures. While recent deep learning models have made significant progress, individual models often exhibit limitations-some may excel at capturing spatial features, while others are better at modeling temporal dependencies or handling class imbalance.MethodsThis study investigates the use of ensemble learning to combine the complementary strengths of diverse architectures, aiming to mitigate individual model weaknesses and improve performance in surgical phase recognition using the Cholec80 dataset. A variety of advanced deep learning architectures was integrated into a single ensemble. Models were carefully selected and tuned to ensure diversity, resulting in a final set of 15 unique ensembles. Ensemble strategies were explored to determine the most effective method for combining the distinct models.ResultsThe results demonstrated that ensemble learning significantly improved performance. Among the ensemble strategies tested, majority voting achieved the highest F1-score, followed by the proposed artificial neural network StackingNet. Ensembles with high model diversity showed superior performance compared to those with lower diversity. The optimal ensemble configuration integrated top-performing models from different architectures, leading to improvements in accuracy, F1-score, and Jaccard Index by 1.48 %, 3.68 %, and 5.43 %, respectively, compared to the best individual models.ConclusionThis study demonstrates that ensemble learning can substantially enhance surgical phase recognition by leveraging the complementary strengths of diverse deep learning models. Ensemble size, diversity, and meta-model selection were identified as key factors influencing performance. The resulting improvements translate into clinically meaningful benefits by enabling more reliable context-aware guidance, reducing misclassifications during critical phases, and improving surgeons' trust in artificial intelligence (AI) systems.

Automatic recognition of surgical phases plays a critical role in enabling intelligent, context-aware support systems during operative procedures. The inherent variability of surgical techniques and intraoperative conditions makes precise surgical phase recognition (SPR) a challenging task. This study explores ensemble learning as a strategy to improve phase recognition performance using a learnable fusion approach. A selection of state-of-the-art deep learning models was trained and tuned to capture complementary aspects of the task. This resulted in 14 base models with varied backbones and parameter settings. To aggregate the output probabilities of the base models, a lightweight fully connected network, referred to as StackingNet, was designed as a metamodel capable of learning to generate final predictions from their outputs. This approach outperformed the best individual base model within the respective ensemble in 14 out of 15 ensemble configurations, achieving a maximum F1-score improvement of 3.3 %. These results demonstrate that learnable ensemble fusion can significantly enhance accuracy of surgical phase recognition, highlighting its potential in the development of intelligent surgical assistance systems.

Elastic structures from polymer are a promising alternative to conventional metal springs under challenging conditions like in magnet resonance imaging (MRI) environments, where ferromagnetic materials cannot be used. The suitability of additive manufacturing to manufacture polymeric springs was assessed within the paper with a dedicated focus on stiffness reproducibility and fatigue behavior. Multi-Jet Fusion (MJF), Selective Laser Sintering (SLS), and Fused Deposition Modeling (FDM) were evaluated as manufacturing technologies. Three spring designs were conceived based on a heuristic approach, taking into account the constraint of anisotropic material behavior. MJF and SLS were used to print all three designs, respectively. Use of FDM was limited to print one design, the others were not appropriate for FDM due to high complexity. The anisotropy in mechanical characteristics of SLS and MJF printing technologies was assessed to estimate its possible influence on spring stiffness. MJF-printed tensile specimens show more anisotropic material behavior compared to SLS-printed ones. Overall suitability of FDM to print springs was shown to be limited due to design constraints and manufacturing limitations like warping and in-plane delaminations between deposited polymer strands, as well as very limited applicable cycles during fatigue testing. MJF-printed springs showed higher variability in geometric dimensions compared to SLS. Slight variances in geometric dimensions were shown to crucially influence spring stiffness, thus lowering the reliability of the MJF technology for reproducible springs. Fatigue life of either SLS or MJF samples was shown to be appropriate as all springs survived 100,000 load cycles with moderate loss of reaction force below 16

One of the heavily burdened actors in the operating theatre is the scrub nurse who is responsible for the organized and orderly workflow in the operating theatre. One of the main tasks of the scrub nurse is to perform quick and appropriate instrumentation because fast and proactive instrumentation is crucial for the success of the operation. The actual deficiency of specialist staff in the healthcare sector is motivating the automation of repetitive tasks and simple work steps in order to relieve staff in the future and free up capacity for complex and important activities. A robotic assistance system appears to be a suitable solution for many of these challenges. The vision is to develop advanced assistance systems for the handling of instruments in future surgical procedures. In this work, surgery of the carpal tunnel was chosen as example task. The focus lied on the type and scope of the individual working steps. The analysis was carried out by video analyses and by observing corresponding surgical procedures. Based on the results of the analysis, potentials for improving the workflow were derived and requirements for a robotic assistance system were defined.